[SciPy-dev] scipy_distutils/*_info.py files

eric eric at scipy.org
Mon Feb 18 11:53:39 EST 2002


----- Original Message -----
From: "Pearu Peterson" <pearu at cens.ioc.ee>
To: <scipy-dev at scipy.org>
Sent: Sunday, February 17, 2002 6:48 PM
Subject: [SciPy-dev] scipy_distutils/*_info.py files


>
> Hi,
>
> I have introduced scipy_distutils/x11_info.py that similarly to atlas_info
> determines the locations of X11 libraries and header files.
> This should fix the long running setup_xplt.py issue. Let me know if I
> missed something.
>
> Note also that x11_info.get_info returns a dictionary that contain items
> like libraries:list,library_dirs:list,etc (this is different from
> the current atlas_info.get_atlas_info).

I like your approach better.  It also mirror how the configuration() functions
in setup_xxx.py work.  Lets plan on returning dictionaries from now on, and also
fixing the atlas_info module.

>
> Seems like that there will be a collection of *_info.py files. I wonder
> what would be the right place to keep these files. Currently they are in
> scipy_distutils though scipy_distutils itself does not use them at
> all.

Yeah, they went there originally because they were part of the setup process.
Also, they seem useful to other people trying to build modules.  For example,
other developers building modules that use LAPACK can use the atlas_info file.
This makes me think they should all live in a central place so that people know
where to look for them.  scipy_distutils seems as good of a place as any.  I
guess we could make a scipy_distutils.config sub-package directory.  Maybe that
would be better.?
>
> It is not a very important matter but may be somebody (Eric ?) had a
> vision on this (unified specification,etc) that we should try to follow.

Not very important to me.  I just want them somewhere easily accessible to the
setup_xxx scripts.  I don't think they should be in a sub-package of SciPy
because this presents problems during the install phase of SciPy.

If anyone else has a grand plan (that doesn't include re-writing autoconf in
Python...) lets talk about it.

eric






More information about the SciPy-Dev mailing list